RUAIH certification requirements — what is known
Start with the honest sentence most pages on this subject omit: the full RUAIH standards text is not freely public. The certification launched on 1 June 2026 and a standards manual exists — organizations are certifying against it — but the complete text is obtained through The Joint Commission, and any article listing “the RUAIH requirements” as a settled checklist without that access is reciting inference as text. This page separates the published record from the inference.
What the published record supports
The certification’s launch names five standards areas, which map directly onto the September 2025 Joint Commission–CHAI guidance:
- Governance — an accountable structure: policy, committee, inventory, executive reporting, resourcing.
- Effective data management — how data enters and leaves AI systems, minimum-necessary use, security controls.
- Risk and bias reduction — risk tiering at intake, bias and equity assessment, vendor due diligence.
- Monitoring, evaluating and validating — local validation before deployment, performance monitoring against thresholds, safety event reporting.
- Transparency, education and training — patient disclosure and role-specific workforce training.
The manual’s published eligibility criteria add texture: applicants must already have a governance structure for AI oversight and established processes for risk and bias assessment, data security and use protections, ongoing quality monitoring, voluntary reporting of AI safety events, patient privacy and transparency, and staff education and training — in other words, the certification examines a program that exists, not a plan to build one. Joint Commission accreditation is not required; compliance with applicable federal requirements (such as the CMS Conditions of Participation) qualifies.
Each area breaks into working controls — the Institute’s fifteen-control crosswalk maps every one to its CHAI playbook and NIST AI RMF function, with the evidence artifact an assessor would ask to see.
What is still not public
Scoring thresholds. Survey mechanics. Documentation formats. Renewal cycles. Fees. None of it has been published openly, and this page will update when it is — the review date above is the promise.
The planning posture
Prepare against the published record and the consensus it confirms. The evidence divides into what can be drafted (charters, policies, procedures — a week each) and what can only accumulate (monitoring history, training records, committee minutes). The preparation guide orders the work by which clocks cannot be started retroactively.
Asked alongside this
Can I see the official RUAIH standards?
The standards manual exists — the certification launched 1 June 2026 and its eligibility criteria have circulated publicly — but the full standards text is obtained through The Joint Commission rather than published freely. The five standards areas were named in the launch announcement and trace directly to the September 2025 Joint Commission–CHAI guidance.
What evidence should we assemble before engaging with the manual?
The evidence that cannot be produced quickly: a maintained AI inventory, a chartered governance committee with minutes, performance monitoring records against locked baselines, and training completion records. Documents can be drafted in a week; a monitoring history can only accumulate.
Do we need Joint Commission accreditation to apply?
No. The published eligibility criteria require a U.S. organization that is Joint Commission–accredited OR compliant with applicable federal requirements such as the CMS Conditions of Participation, plus an established AI governance structure and processes for risk and bias assessment, data protections, quality monitoring, safety-event reporting, transparency and training.
← All RUAIH questions, areas and terms · The complete healthcare AI governance guide · Score your readiness